Credo at Goldman Sachs conference: optical ramp drives bigger vision

Credo raised fiscal 2027 revenue-growth guidance to 85% year over year, maintained its $600 million optics-revenue target, and forecast sequential growth accelerating from 11% in Q2 to 20% in Q3 and 30% in Q4. Management expects gross margin to hold near 68% despite a larger optical mix and said a $1 billion quarterly revenue run rate could be achievable in early fiscal 2028. The company is expanding from its core AI data-center copper connectivity business into differentiated optical products, targeting hundreds of thousands of optical units per month by fiscal year-end 2027, while flagging customer deployment timing and execution as key risks.

Analysis

CRDO’s investment case is shifting from an AI-capex beta into a qualification-and-reliability bottleneck story. If its diagnostic layer becomes embedded in cluster commissioning workflows, switching costs rise materially: customers are not merely replacing a cable or module, but risking utilization losses across expensive GPU fleets. That supports premium ASPs and makes the key competitive comparison less with commodity optics vendors than with ANET, CSCO and in-house hyperscaler engineering teams that could attempt to standardize telemetry at the switch or NIC layer.

The near-term setup is vulnerable to expectations rather than demand. A sharply back-end-loaded growth profile leaves little room for supply-chain delays, qualification slippage, or a single hyperscaler pause; the first evidence of mix-driven gross-margin deterioration would challenge the claim that optical can scale without dilution. Higher real-rate pressure is also a meaningful second-order risk: CRDO is a long-duration AI growth asset, so multiple compression can overwhelm upward earnings revisions over days to weeks even if operational execution remains intact.

Over the next 1-3 months, the important catalyst is independent customer validation of the reliability advantage and evidence that production yields can support the planned optical ramp. Over 6-18 months, the upside rests on whether the company converts bespoke deployments into repeatable platforms across cloud and neocloud customers; that would expand revenue per GPU cluster and reduce dependence on its incumbent short-reach product cycle. Consensus may be underestimating this platform optionality, but it may also be over-crediting management’s margin confidence before unit economics, customer concentration, and acquisition integration are disclosed.

Key Decisions for Investors

• Maintain/establish a measured long CRDO only on broad AI-networking or rate-driven weakness, rather than chasing conference momentum; size as a 6-12 month position with a 2:1 upside/downside target. Thesis invalidation: a material reduction in the next two quarters’ sequential-growth path, gross margin below the stated mid/high-60s range, or evidence that a top customer is pausing deployment.
• Express relative preference through long CRDO / short CSCO in equal dollar beta-adjusted size over 3-6 months. The trade isolates AI-cluster connectivity spending and premium reliability exposure against enterprise/networking demand that remains more cyclical; close if CSCO’s AI networking order growth materially accelerates or CRDO reports optical-margin dilution.
• Do not underwrite the fiscal-2028 micro-emitter opportunity yet. Set an event-driven alert around the upcoming industry demonstration and subsequent customer qualification disclosures; upgrade only after named design wins, production-yield data, and volume commitments make the TAM claim independently testable.
• For existing CRDO longs, hedge 1-3 month macro duration risk with a partial short in QQQ or SOXX around inflation/Fed catalysts rather than selling the single-name position. The hedge is warranted while valuation and forward revenue multiple data are absent; remove it if rate pressure eases and CRDO’s order visibility broadens beyond concentrated customers.